Google TPUv7 Delivers 1.67x Performance per Dollar over GB300 in SemiAnalysis Benchmark

Google submitted TPUv7 FP8 with vLLM to SemiAnalysis's InferenceX benchmark for the first time, achieving 1.67x performance per dollar over NVIDIA's GB300 NVL72, with external TCO as low as $0.181 per million tokens.

2026-10-14 ~ 2026-10-15 · 2 related posts

• SemiAnalysis teases TPUv7 TCO data: $0.181/M tokens vs $0.222 B200 — firstadopter · 2026-10-14
• Google's TPUv7 Posts 1.67x Better Perf per Dollar Than GB300 NVL72 in SemiAnalysis InferenceX — scaling01 · 2026-10-15